Journal of Paleolimnology (2005) 34: 31–49
DOI 10.1007/s10933-005-2396-2
Springer 2005
Paleolimnological investigations of anthropogenic environmental change
in Lake Tanganyika: III. Physical stratigraphy and charcoal analysis
Manuel R. Palacios-Fest1,*, Andrew S. Cohen2, Kiram Lezzar2, Louis Nahimana3 and
Brandon M. Tanner2
1
Terra Nostra, Tucson, AZ 85741, USA; 2Department of Geosciences, University of Arizona, Tucson, AZ
85721, USA; 3De´partement des Science de la Terre, Universite´ de Burundi Bujumbura, Burundi 2700; *Author
for correspondence (e-mail: terra_nostra_mx@yahoo.com.mx)
Received 10 July 2004; accepted in revised form 15 January 2005
Key words: Catchment deforestation, Charcoal, Lake Tanganyika, Sedimentation rates, Soil erosion
Abstract
Documenting the history of catchment deforestation using paleolimnological data involves understanding
both the timing and magnitude of change in the input of erosional products to the downstream lake. These
products include both physically-eroded soil and the byproducts of burning, primarily charcoal, which arise
from both intentional and climatically-induced changes in fire frequency. As a part of the Lake Tanganyika
Biodiversity Project’s special study on sedimentation, we have investigated the sedimentological composition of seven dated cores from six deltas or delta complexes along the east coast of Lake Tanganyika: the
Lubulungu River delta, the Kabesi River delta, the Nyasanga/Kahama River delta, and the Mwamgongo
River delta in Tanzania, and the Nyamusenyi River delta and Karonge/Kirasa River delta in Burundi.
Changes in sediment mass accumulation rates, composition, and charcoal flux in the littoral and sublittoral
zones of the lake that can be linked to watershed disturbance factors in the deltas were examined. Total
organic carbon accumulation rates, in particular, are strongly linked to higher sediment mass accumulation
from terrestrial sources, and show striking mid-20th century increases at disturbed watershed deltas that
may indicate a connection between increased watershed erosion and increased nearshore productivity.
However, changes in sedimentation patterns are not solely correlated with the 20th century period of
increasing human population in the basin. Fire activity, as recorded by charcoal accumulation rates, was
also elevated during arid intervals of the 13th–early 19th centuries. Some differences between northern and
southern sedimentation histories appear to be correlated with different histories of human population in
central Tanzania in contrast with northern Tanzania and Burundi.
Introduction
Understanding the impact of watershed deforestation on lake ecosystems requires the determination of changes in both the quality and quantity of
sediment being discharged into the lake. The
stratigraphic analysis of well-dated cores from
deltaic regions of lakes, where sediments are initially discharged, provides critical data for interpreting the timing and magnitude of these impacts
(e.g., deforestation, change in sediment composition in the lake). In large lakes, where human
32
impacts vary greatly between influent watersheds,
paleolimnologic studies can also provide a means
of comparing sedimentologic responses of
watershed deforestation across a spectrum of preexisting watershed characteristics and impact levels. For example, a combination of watershed size
and magnitude of deforestation may be responsible for changes in sediment inundation that in turn
will impact the ecosystem affecting life patterns
(e.g., species ecological replacement).
Here, we describe the physical stratigraphy of
cores collected by the Lake Tanganyika Biodiversity Project’s Special Study on Sedimentation
Impacts. A companion paper by Cohen et al.
(2005a) describes the background and rationale for
this study, and provides location maps, site characteristics and coring techniques used to obtain the
cores described here. Another companion paper
describes the geochronology and age models used
here (McKee et al. 2005). Briefly, the cores
described here were collected from a series of river
deltas along the eastern margin of Lake
Tanganyika, which span a spectrum of watershed
disturbance and size characteristics. These deltas
lie offshore from the following rivers (in order
from south to north): the Lubulungu River (low
disturbance, small-sized drainage area: 50 km2,
central Tanzanian coastline), Kabesi River (medium disturbance, medium-sized drainage area:
120 km2, central Tanzanian coastline), Nyasanga/
Kahama Rivers (low disturbance, very small-sized
drainage area: 3.8 km2, northern Tanzanian
coastline) and Mwamgongo River (high disturbance, very small-sized drainage area: 7.7 km2,
northern Tanzanian coastline), Nyamusenyi River
(extremely high disturbance, small-sized drainage
area: 30 km2, northern Burundi coastline), and
Karonge/Kirasa Rivers (extremely high disturbance, medium-sized drainage area-combined area
162 km2, northern Burundi coastline (see
Figures 1–5 and Table 1 of Cohen et al. 2005a).
Watersheds were characterized as currently
experiencing low, medium or high levels of disturbance based on the proportion of mature forest/woodland cover existing in the watershed. Low
disturbance areas have forest/woodland cover in
both the delta plain and upland portions of the
watershed, with extremely limited or no agricultural/grazing activity. Medium disturbance watersheds have extensive agricultural development in
the lowland and/or delta plain regions of the
watershed, but retain forest, woodland, or mixed
woodland/grassland cover in their uplands, with
the entire watershed retaining between 25 and
75% forest/woodland cover. High disturbance
areas are regions where >75% forest/woodland
cover has been removed. The two deltas in
Burundi are additionally characterized as ‘extremely highly’ disturbed because these watersheds
have undergone extensive surface slope failure.
Materials and methods
Numerous multicores were collected at each delta
during this study, of which only a small number
could be analyzed in detail. We selected the best
cores for use in this study, based on a combination
of likelihood of providing continuous records,
quality and quantity of indicator materials, and
comparability of coring stations vis-a-vis their distances from shore and water depths. Three splits
from each core were sampled, normally at 3-cm
intervals, for loss-on-ignition, sedimentological
(granulometric and micropaleontologic) and palynological analysis. The 3-cm sampling interval used
corresponds to the mean depth of bioturbation and
sample time averaging observed in X-radiographs
of deltaic cores from Lake Tanganyika. A total of
109 samples were prepared for each analysis.
For this exploratory investigation, a simple set
of informative and inexpensive indicators was
chosen. Granulometric analysis (grain size) provides indications of significant changes in the
nature of eroded materials within a watershed and
the strength of sediment delivery systems to the
coring site. Because total carbonate content in
most cores was low (<1%), the contribution of
non-terrigenous sources of coarser-grained particles in these cores is probably negligible. Water
content and bulk density were measured as supporting data for interpreting sedimentation rates.
Total carbonate content and total organic matter
were measured as probable indicators of benthic
secondary productivity, terrestrial/aquatic productivity and organic matter flux.
The full age spectral data, associated probabilities, and standard errors for each age date, along
with discussion of occasional discrepancies between
14
C and 210Pb age dates, age models, and sediment
accumulation rate estimates used in this study are
discussed elsewhere (McKee et al. 2005). The 14C
33
dates presented in this paper are calendar-year age
estimates (age B.P. in parentheses, relative to
1950 A.D.) based on the median values of the 14C
probability spectra derived from each uncalibrated
age date. ‘Ultramodern’ (i.e., post-bomb) 14C dates,
in conjunction with 210Pb age models constrain the
ages of the upper parts of most of the cores. Sediment accumulation rate estimates are based on
210
Pb profiles except for those intervals that predate
the 210Pb time scale (last 150 years). Where significant discrepancies exist between 210Pb and 14C age
estimates we have favored the former, because of the
well known uncertainties in 14C geochronology for
the last few hundred years.
All sedimentologic splits were prepared for
granulometric and micropaleontologic analysis
using the USGS freeze–thaw technique (Forester
1991), modified by Palacios-Fest (1994). Approximately 10–20 g wet weight was used for each
sample. For granulometric analysis, samples were
wet-sieved in a set of three sieves of >1 mm,
>106 lm and >63 lm mesh sizes to obtain the
coarse, medium and fine sand fractions.
All samples for loss-on-ignition (LOI) were prepared using the technique described in Bengtsson
and Enell (1986) and Boyle (2004). This involves
step-wise heating and reweighing of samples to 105,
550, and 925 C for the determination of water
content, organic matter LOI [0.40 · conversion
factor for total organic carbon (TOC)], and inorganic matter LOI [0.12 · conversion factor for total
inorganic carbon (TIC)], respectively. Data are presented as both abundance per gram and as a flux
(mg cm 2 yr 1) based on component proportions
and mass accumulation rates, and are plotted against
time for easier comparison between core sites.
Charcoal abundance was estimated as the number of charred particles per gram retained on the
>106 lm sieve fraction only, where most of the
volume was concentrated, as an indicator of total
abundance. This coarser charcoal fraction provides
a good indication of local, within-watershed production, as opposed to long-distance aerial transport. We calculated abundance both as number per
gram and as a flux (# cm 2 yr 1).
Results
[Note: cores are described from South to North.
For details on core sites, see Cohen et al. (2005a)]
Lubulungu delta cores, LT-98-2M and LT-98-12M,
central Mahale Mountains region, Tanzania (low
disturbance, small-sized drainage area)
Two cores were collected from the west-central
part of the Lubulungu River delta, LT-98-2M and
LT-98-12M. Core LT-98-2M was collected in
110 m of water depth in the central plain of the
delta, about 1.5 km offshore and west of the
Mahale Mountains National Park, Tanzania.
Swimming copepods and clear water at the sediment/water interface indicate that the sediment
surface was undisturbed during collection. However, 14C geochronologic data, discussed in
McKee et al. (2005) suggests that this core site was
an area of either non-deposition or erosion over
the last few hundred years. The core consists of
49 cm of alternating massive sandy clay and clay
(partly laminated in the lower 10 cm), either with
shell fragments or plant remains, or occasionally
both (Figure 1). A notable fining upwards of
sediments occurs above 35 cm, about 700–
600 B.C. (2650–2550 B.P.) (Figure 2). A secondary fining event occurs higher in the core, above
about 25 cm. Assuming continuous sedimentation, this would correspond to about 400–
500 A.D. (1550–1450 B.P.), although it is possible
that substantial hiatuses exist in this record, given
these fining trends and the long interval covered
by the core. Both TOC and TIC measurements
show increases in the LT-98-2M core. Concentration and mass accumulation rates change, with
notable increases in TOC and TIC mass accumulation rates starting about 100 B.C. (2050 B.P.),
shortly before the first appearance of calcareous
fossils in the core. These trends correlate with
upcore granulometric changes to finer textures.
Total organic carbon (TOC) values remain high
(5%) throughout the rest of the core, whereas
TIC values (reflecting mollusc fragments, and to a
lesser extent, ostracode concentrations) decline
after 1100 A.D. (850 B.P.). Both granulometric and geochronologic data suggest that this
change in TIC is probably a result of initial
increases in calcium carbonate availability and/or
preservation and later productivity of carbonateproducing organisms, followed by siliciclastic
dilution caused by higher rates of mud input.
Charcoal concentration is extremely low (<5
particles cm 2 yr 1) before 1 A.D. (1950 B.P.).
Between 1 and 100 A.D. (1950–1850 B.P.),
34
Figure 1. Lithostratigraphy of core LT-98-2M, central Lubulungu delta, 110 m water depth. Total core length 49 cm. See Cohen et al.
(2005a, Figure 2) for location and bathymetric map.
accumulation rates increased in the upper part of
core LT-98-2M. Charcoal increase coincided with
the first occurrences of ostracodes and molluscs in
the core, and greatly accelerating after 1200–
1300 A.D. (750–650 B.P.). Higher charcoal concentrations are evident from 13th to early 19th
centuries in this and several of our other core
records (cores LT-98-12M, LT-98-18M and LT98-58M). Charcoal concentrations probably
reflect regional aridity and increased fire activity,
in accord with other records showing periods of
extremely arid conditions and low lake levels
during portions of the Little Ice Age (Cohen et al.
1997; Nicholson 1999; Verschuren et al. 2000; Alin
et al. 2002; Alin and Cohen 2003).
Core LT-98-12M was collected in 126-m water
depth, about 500 m northeast of core LT-98-2M
(1.2 km from shore), on a narrower and deeper
portion of the delta front. Unlike LT-98-2M, the
top of core LT-98-12M appears to be ‘modern’,
although 19–20th century sediment accumulation
rates at this site have been very low, and the
35
Figure 2. Sedimentologic and charcoal profiles for core LT-98-2M.
temporal sampling resolution as a result was not as
good as in other intervals. Core LT-98-12M consists of 40 cm of massive sandy clay, alternating
with either mollusc fragments or plant debris or
wood fragments, or occasionally both (Figure 3).
In general, sediments coarsen upwards, mostly
through the upper 10 cm (18–20th centuries)
(Figure 4). Low concentrations of sand occur
throughout the lower portion of LT-98-12M, with
generally higher values evident after the early 18th
century. LT-98-12M displays high TOC values and
low TIC values throughout the core. Both vary
only slightly, in ways that are not evidently correlated with other variables. Total organic carbon
(TOC) and TIC MARs decline dramatically at the
top of the core as a consequence of overall steep
declines in sedimentation rates since the late 18th
century. A decrease in charcoal abundance and
major decline in accumulation rate occurs above
30 cm (early 16th century?), with low values persisting up to 15 cm (late 17th–early 18th century).
This is followed by a rise in charcoal concentration
in the upper part of the core. However, this
increase appears to be largely an artifact of the low
background sedimentation rates; when calculated
as an accumulation rate charcoal flux shows a
secondary rise in the mid-late 18th century
followed by declines to very low levels at the top of
the core.
Kabesi delta core, LT-98-18M, north Mahale
Mountains region, Tanzania (intermediate
disturbance, medium-sized drainage area)
Core LT-98-18M was collected about 1.5 km offshore of the Kabesi River mouth, in 75-m water
depth. The core top contained a live gastropod
(Paramelania iridescens) in living position, indicating perfect recovery of the sediment–water
interface and oxic conditions at the core site. The
core consists of 42 cm of brown massive mud
(Figure 5), which display an increasing proportion
of sand in the uppermost 12 cm of the core (from
<0.95 to 2.48%), coincident with a 3- to 4-fold
increase in sedimentation rates dating from the
early 1960s (Figure 6). Total organic carbon
(TOC) concentrations range between 4 and 5%,
and decline gradually but systematically above
20 cm, 1899. However, TOC accumulation rates
rise dramatically after the early 1960s, in concert
with overall increasing sedimentation rates. Total
inorganic carbon (TIC) content increases slightly
at the same time, but is low throughout the core.
36
Figure 3. Lithostratigraphy of core LT-98-12M, central Lubulungu delta, 126 m water depth. Total core length 40 cm. See Cohen
et al. (2005a, Figure 2) for location and bathymetric map.
As with the TOC record, the relatively uniform
TIC concentration throughout the core actually
represents a major increase in accumulation rates
of calcium carbonate over the past 40 years at
this site. For TIC, this probably represents in situ
carbonate production, since there is no source of
CaCO3-particulate matter in the Kabesi
watershed. In the case of TOC, the increase may
reflect increased aquatic organic matter accumulation,
increased terrestrial organic inputs, or some combination of the two. LT-98-18M contains relatively
low concentrations of charcoal, and accumulation
rates are also low. Charcoal concentrations and
accumulation rates decline fairly continuously
from the base of the core (mid-18th century) until
the mid-20th century, at which time they begin to
rise again, followed by an abrupt decline in the late
1980s.
37
Figure 4. Sedimentologic and charcoal profiles for core LT-98-12M. *Available sample material from the 1910 horizon was too small
to obtain accurate TOC and TIC measurements.
Nyasanga/Kahama deltas core, LT-98-58M,
northern Tanzania (low disturbance, very
small-sized drainage area)
Core LT-98-58M was collected in 76-m water
depth, about 300 m offshore from Gombe Stream
National Park and the Nyasanga/Kahama coastal
sand belt. The precise coring site was located on a
topographic bench on an otherwise steep slope,
thus allowing for fine sediment accumulation. The
core consists of 39 cm of brownish to dark gray
clays (Figure 7). Sediments alternate between
laminated silty clay and massive clay with carbonate layers and shell fragments, and display a
slight coarsening upwards above 16 cm (1898)
(Figure 8). LT-98-58M displays moderately high
(2.5–4%) TOC levels throughout the core, with no
systematic trends evident in either concentration
or accumulation rate. Total inorganic carbon
(TIC) concentration is relatively high compared to
other cores, although measured as an accumulation rate it is intermediate. Total inorganic carbon
(TIC) is also very variable throughout the core,
with higher values in the mid-18th century, a drop
in the late 18th century and then a marked rise in
the early 18th century. Following this period, there
was marked decline in the mid-19th century followed by a significant rise after 1920 in both
concentration and accumulation rate. Charcoal
abundance is extremely high in the LT-98-58M
core and rises to extraordinary levels at the core
top, with the highest flux rates observed in any
core studied. This result is unexpected, as the area
is protected from intentional burning today (high
values occur both pre- and post-1968, the date of
National Park establishment). Very high values
occur in samples from both the late 18th/early 19th
century, and the late 20th century. Much lower
concentrations and accumulation rates occur from
the mid-19th to mid-20th century, although these
values are still high in comparison with other core
localities. The high values of the late 20th century
suggest that charcoal must be transported by flotation over distances exceeding that separating this
study site from the park boundary (i.e., several
kilometers), or alternately, that wildfire within the
park is for some reason anomalously high. We
consider the latter explanation unlikely, because
the difference between park and ‘non-park’ in
terms of watershed land usage and seasonal
38
Figure 5. Lithostratigraphy of core LT-98-18M, Kabesi River delta, 75 m water depth. Total core length 42 cm. See Cohen et al.
(2005a, Figure 3) for location and bathymetric map.
burning is dramatic in this area. This signal of
sediment input for floating fractions from areas
outside of the immediate vicinity of this small delta
is also evident in the pollen record of this site,
discussed in the companion paper by Msaky et al.
(2005).
Mwamgongo delta core, LT-98-37M northern
Tanzania (high disturbance, very small-sized
drainage area)
Core LT-98-37M was collected in 95-m water
depth about 300 m offshore from the Mwamgongo
39
Figure 6. Sedimentologic and charcoal profiles for core LT-98-18M.
River, north of Gombe Stream National Park,
Tanzania. Like the Nyasanga/Kahama site, this
coring location was a flat bench on an otherwise
steep slope. The overall core condition and core
top sediment–water interface preservation were
excellent. The sediment/water interface consists of
flocculent clay with abundant live copepods,
ostracodes and snails. The core consists of 45 cm
of brownish clays (Figure 9). Alternating massive
silty clay and dark sandy organic clay occur
throughout, along with carbonates, and the core
displays a slight fining upwards. A striking transition to reddish clays near the core top was
observed in this core and the other cores collected
from the Mwamgongo delta. Similar color changes
occur at the tops of cores from other disturbed
Tanzanian and Burundian delta cores, and probably indicate high rates of eroded, lateritic soil
accumulation, which accumulated so quickly that
the reduction interface is well below the sediment–
water interface. An upcore decrease in the proportion of coarse sand occurs in LT-98-37M, with
pronounced declines near the base of the core
(15th century) and then gradually through the
early 19th century, after which sand content rises
slightly (Figure 10). This grain size trend closely
parallels sediment mass accumulation rates (coarser sediments = higher rates). LT-98-37M displays high values of TOC throughout the core
(4% average, with one sample from the early
1970s reaching nearly 7%). Total organic carbon
(TOC) accumulation rates are moderate compared
with other cores in the lower part of the core (15–
16th centuries) then decline to low levels and rise
again to intermediate levels during the 19th century. LT-98-37M also displays some of the highest
TIC values seen in any core (2–4%). Their pattern
of higher concentrations and accumulation rates in
the lower part of the core, dropping to low levels in
the middle of the core, and then rising again in the
upper part of the core mirrors TOC, except that
the dramatic rise in TIC (late 19th century) and its
subsequent decline (1940s) occur somewhat earlier
for TIC than TOC. Moderately high values of
charcoal concentration and accumulation rates
exist throughout the core. Charcoal accumulation
is relatively high prior to the mid-16th century,
then declines to much lower levels through the
18th century, and rises again during the late 19–
20th centuries.
40
Figure 7. Lithostratigraphy of core LT-98-58M, Nyasanga/Kahama Rivers delta, 76 m water depth. Total core length 39 cm. See
(Cohen et al. 2005a, Figure 4) for location and bathymetric map.
Nyamusenyi delta core, LT-98-98M, northern
Burundi (high disturbance, small-sized
drainage area)
Core LT-98-98M was collected in 60-m water
depth, about 100 m north of the Nyamuseni River
discharge point and approximately 200 m offshore. The core site is on the proximal part of an
elongate spur or ridge extending off the delta,
which, based on its shape, may be a small structural high. The core and core top were in excellent
condition at recovery. The core consists of 37 cm
of alternating brown sandy clay and micaceous
clay, with or without plant debris (Figure 11). The
sediment–water interface consists of clayey sand.
Apart from a transition from sand to claydominated sediment at the base of the core, there
are no clear grain-size trends through the core
(Figure 12). Core LT-98-98M displays relatively
constant and moderately high concentrations (2–
4%) of TOC throughout the core, coupled with
consistently low TIC concentrations (reflected in
the near absence of shelly fossils in this core).
Abundant root fragments near the base of the core
(late 1950s–early 1960s) indicate the rapid accumulation of reworked soils, consistent with the
41
Figure 8. Sedimentologic and charcoal profiles for core LT-98-58M.
inference of increased soil erosion rates starting at
that time discussed elsewhere (McKee et al. 2005).
Accumulation rates of both TOC and TIC show
an initial rise in the early 1960s, then remain relatively stable through the early 1990s, and then
increased dramatically after the early 1990s. Small,
needle-shaped rosettes of aragonite, formed in the
water column, are extremely abundant in the surficial sediments of this area and probably represent
the bulk of the TIC observed in these sediments.
The increased flux in this aragonite may reflect
increasing primary productivity rates in surface
waters accompanying the increasing sediment
input, a phenomenon observed at the mouths of
other rivers in the Tanganyika basin that discharge
large volumes of nutrients (Castañeda et al. 1999).
Based on sedimentation rate changes, this nutrient
enrichment was probably underway after the early
1960s, and increased dramatically in the 1990s.
Core LT-98-98M displays relatively constant
charcoal concentrations and gradually rising
accumulation rates prior to the early 1980s. Peak
charcoal concentrations and accumulation rate in
the early 1980s were followed by a dramatic
decline after the 1980s, probably reflect the
declining availability of forest cover for burning
within this watershed, as conversion to agricultural and disturbed lands was more or less completed.
Karonge/Kirasa deltas core, LT-98-82M, northern
Burundi (high disturbance, medium-sized
drainage area)
Core LT-98-82M was collected in 96-m water
depth, 1.2 km west of the Karonge and Kirasa
delta, on a broad, gently dipping slope. The core
and core top were both in excellent condition at
recovery. The core consists of 46 cm of alternating
dark gray and gray, laminated or massive clays
(Figure 13). The core shows a notable upcore decline in sand concentration, with brief reversals
accompanying periods of increased sedimentation
rates (Figure 14). Relatively constant and high
TOC concentrations (slightly more abundant
proportionately than at the nearby LT-98-98M)
occur throughout the core. Total organic carbon
(TOC) accumulation rate changes reflect overall
sediment accumulation rate changes, with a major
increase in TOC MAR after the early 1960s
42
Figure 9. Lithostratigraphy of core LT-98-37M, Mwamgongo River delta, 95 m water depth. Total core length 45 cm. See (Cohen
et al. 2005a, Figure 4) for location and bathymetric map.
reflecting background increases in sedimentation
rates (Figure 14). Total inorganic carbon (TIC)
concentration is also relatively constant throughout the core, with the exception of an interval in
the mid-19th century, when concentration and
accumulation rates rise significantly. As with TOC,
TIC MARs rise again to high levels starting in the
early 1960s.
As with LT-98-98M, abundant visible terrestrial plant debris throughout the upper part of
LT-98-82M suggests that the TOC increase is
primarily driven by allochthonous organic matter
coming from terrestrial sources. Somewhat surprisingly, given the highly disturbed nature of this
watershed, core LT-98-82M contains a relatively
low abundance of charcoal throughout much of
43
Figure 10. Sedimentologic and charcoal profiles for core LT-98-37M.
the core, mirroring low accumulation rates,
except for two levels (36 and 27 cm = early-mid19th century) and towards the core top (late 20th
century).
Intersite comparisons and discussion
Total organic carbon (TOC) values ranged
between 3–7% in all cores, with the notable
exception of the lowermost portion of LT-98-2M
(pre 500 B.C./2450 B.P.), which was sandy
and TOC-poor. This interval, unrecorded in any
other cores, may have been a period of substantially lower lake levels. This evidence for lower
lake levels in the LT-98-2M core is consistent
with earlier studies (Haberyan and Hecky 1987;
Casanova and Hillaire-Marcel 1992; Cohen et al.
1997), all of which suggest a probable closure of
the Ruzizi River outflow from Lake Kivu to Lake
Tanganyika for an extended portion of the Late
Holocene. These earlier studies suggested that the
rise in Lake Tanganyika water level that accompanied the opening of Lake Kivu occurred at
some time between 100 and 800 A.D. (1850–
1150 B.P.). However, the results from this study
imply that at least some outflow began earlier,
between 1000 and 600 B.C. (2950–2550 B.P.). In
large part, these discrepancies probably result
from our age model uncertainties for the lake’s
paleohydrologic history during the 1000 B.C.
(2950 B.P.) and 1000 A.D. (950 B.P.) interval.
Major paleolimnologic changes consistent with
erratically rising lake levels, beginning prior to
620 B.C. (2570 B.P.) (this study) and largely
completed by 580 A.D. (1370 B.P.) (Alin and
Cohen 2003) suggest that inflow from Lake
Kivu probably increased incrementally during
this time.
Systematic increases in TOC accumulation rates
are evident from all of the highly disturbed
watershed cores. This increase occurs abruptly in
the 1960s in the Kabesi River core from central
Tanzania (LT-98-18M). At the more northerly,
disturbed sites with multi-century records (LT-9837M and LT-98-82M) this 1960s rise is also evident, but is preceded by a more gradual increase,
starting in the 19th century. In contrast, no such
increase in TOC accumulation rate is evident in
either of the undisturbed core sites, at Lubulungu
(LT-98-12M) or Nyasanga/Kahama (LT-98-58M),
with the LT-98-12M core actually showing a
44
Figure 11. Lithostratigraphy of core LT-98-98M, Nyamusenyi River delta, 60 m water depth. Total core length 37 cm. See Cohen
et al. (2005a, Figure 5) for location and bathymetric map.
significant decline in TOC MARs after the late
18th–early 19th century.
Total inorganic carbon (TIC) concentrations
were very low (<1%) in all cores from the central
Tanzania and Burundi deltas (LT-98-2M, 12M,
18M, 98M and 82M). At site LT-98-2M, the near
absence of TIC sedimentation prior to 1000 B.C.
(2950 B.P.), and its gradual increase through
100 A.D. (1850 B.P.) is consistent with the interpretation of Lake Kivu closure discussed above,
since that lake is the primary source of calcium
carbonate (and dissolved solutes in general) to
modern Lake Tanganyika.
Total inorganic carbon (TIC) concentrations
were substantially higher in both northern
Tanzanian coast cores (LT-98-58M and 37M) than
in the other cores examined. These intersite differences probably reflect overall lower fluxes of
siliciclastic sediments into the areas offshore from
these smaller watersheds, resulting in lower
degrees of dilution of carbonate sedimentation.
Mass accumulation rate changes in TIC through
the core records for the moderately disturbed site
at Kabesi (LT-98-18M) and the very highly disturbed sites in Burundi (LT-98-98M and LT-9882M) for the late 19th and 20th centuries mirror
45
Figure 12. Sedimentologic and charcoal profiles for core LT-98-98M.
the patterns described above for TOC. At these
sites, TIC accumulation rates may be tracking
benthic invertebrate productivity, since most of
this carbonate is shell material. However, the
complicated patterns of TIC MAR at other sites
argue that other factors are probably involved in
this record. An interval of elevated TIC accumulation in the early 19th century is recorded at one
of the Burundi disturbed sites (LT-98-82M) and
one of the northern Tanzanian disturbed sites (LT98-58M) but not the other nearby sites (LT-9898M and LT-98-37M). These changes in carbonate
production are quite localized and were probably
driven by a wide variety of factors (e.g., intensity
of deforestation, land use, erosion, nutrients in
water, local changes in alkalinity and salinity at
the mouth of the river) besides simply local benthic
productivity. Abundant terrestrial plant debris
throughout the upper part of LT-98-98M suggests
the TOC increase is primarily driven by allochthonous organic matter coming from terrestrial
sources. The increase in TIC flux, by contrast, is
almost certainly autochthonous, because there is
no source of particulate calcium carbonate in the
watershed, and given the extremely low abundance
of both molluscs and ostracodes, this rise cannot
represent an increase in benthic consumers.
Charcoal fragment abundances are very high
(103–104 g m 1) throughout most of the cores
(except LT-98-18M and, surprisingly, LT-9882M), consistent with the common uncontrolled
dry season fire activity, the use of fire in land
clearance and the ubiquitous usage of fire for
cooking and charcoal production. Wildfires are
commonly observed in the Tanzanian coastal
woodlands and grasslands today, although in the
more heavily settled regions of Burundi they are
now much less common. In most cases, charcoal
abundance and accumulation rates cannot be
directly or simply correlated with patterns of land
use in the immediate adjacent watersheds. This is
undoubtedly a consequence of the interactive
effects of climate and changing land use. The role
of climate as an important influence on fire frequency (and therefore charcoal abundance) is
evidenced by the high abundances and accumulation rates of charcoal deposition during the relatively arid Little Ice Age period. The major
increase in charcoal accumulation rates starting in
the 12–13th centuries and reaching a maximum in
the 15–16th centuries in LT-98-2M is particularly
noteworthy in this regard. High charcoal abundances and fluxes at the base of LT-98-12M, and
again in the mid-late 18th century, probably also
46
Figure 13. Lithostratigraphy of core LT-98-82M, Karonge/Kirasa River delta, 96 m water depth. Total core length 46 cm. See Cohen
et al. (2005a, Figure 5) for location and bathymetric map.
correlate with intervals of abundant fires related to
aridity, and are contemporaneous with the interval
of maximal aridity inferred from the palynological
portion of this study (Msaky et al. 2005). Msaky et
al. (2005) indicate that aridity of the region began
500 A.D. (1450 B.P.) with the gradual
replacement of forest by open grassland vegetation. These high charcoal abundances mimic the
more clearly anthropogenic effects of the late 20th
century. In an earlier study of late Holocene
paleoclimates in the Lake Tanganyika region, Alin
and Cohen (2003) suggested that arid conditions
and low lake stands occurred in the mid-late 16th
century and again in the mid-18th century. By
contrast, these results place the arid events slightly
earlier, discrepancies that are likely the result of
remaining uncertainty in our radiometric dating
and age models.
47
Figure 14. Sedimentologic and charcoal profiles for core LT-98-82M.
Reduced charcoal accumulation rates during the
19th century at most of our study sites may reflect
a combination of two factors affecting the region
at the time. First, East/Central Africa appears to
have become generally wetter during that period
relative to the early 19th century (Verschuren in
press). Second, human population densities in
much of the western rift valley region during the
mid-19th century appear to have been greatly
reduced, as a result of both the depredations of
slaving activity in this region and disease (Stanley
1878; Bennett 1970; Koponen 1988). It is important to note, however, that the timing of this
reduction is not synchronous between sites. At
Lubulungu, by far the wettest region of any of the
study areas, charcoal accumulation rates were
relatively high in the mid-18th century but then
declined during the late 18th through the mid-19th
century, and have remained low between that time
and the present. It may be significant, in this
regard, that forest cover in this region of westcentral Tanzania appears to have been increasing
during the period of caravan route incursion, when
human populations are presumed to have been on
the decline. At the Kabesi and Nyasanga/Kahama
sites, lower charcoal accumulation rates started in
the early 19th century and persisted until the mid20th century. Further north, at Mwamgongo, low
sampling resolution in the 18th and 19th century
precludes an accurate determination of the timing
of rising charcoal sedimentation, other than
making it clear that this occurred before the late
19th century. In northern Burundi, the decline
occurred later, in the late 19th century. This geographic pattern may be related to the differences in
human demographic history between the central
Lake Tanganyika catchment (strongly influenced
by the effects of 19th century caravan routes) and
the more northerly Burundian catchments, away
from the caravan trade routes, a point discussed in
more detail in Cohen et al. (2005b). However, at
the present time, we do not know how charcoal
input into the sediments of Lake Tanganyika from
cooking fires or charcoal production might differ
from the pattern produced by uncontrolled wildfires or intentionally-set fires for land clearing.
Understanding such differences might help clarify
the complex patterns observed in our cores. We
also do not know the extent to which anomalously
elevated values of charcoal in these profiles may
represent single large fire events, or alternatively,
integrate longer-term changes in charcoal input.
48
Summary
A sedimentologic analysis of cores collected from
the deltas of disturbed and undisturbed watersheds
of various sizes around Lake Tanganyika shows
that patterns of sediment deposition and charcoal
flux are complex, representing an interplay of
anthropogenic, climatic, and transport phenomena. Fairly clear indications of increased mass
accumulation rates and TOC accumulation rates
are evident from disturbed sites, starting in the
19th century at some sites but increasing rapidly
and apparently synchronously in the mid-20th
century. These rate changes are in some cases
accompanied by accumulation rate increases of
autochthonous inorganic carbon deposition, possibly reflecting in situ rises in benthic productivity
since most carbonate is shell material. They are
also accompanied by rising charcoal fluxes at a
number of sites, which are indicative of land
clearing activities. However, the effect of anthropogenic land clearance is mimicked during arid
intervals by higher charcoal deposition rates and/
or by rising sediment accumulation rates. The midlate 19th century was generally a period of low
charcoal flux at the more southerly sites, consistent
with both wetter conditions and relatively low
human population densities. To the north, high
charcoal accumulation rates in the mid-19th century, followed by an abrupt decline to low rates,
which persisted from the late 19th to mid-20th
century, indicate a different fire and human
demographic history.
Acknowledgements
We thank the United Nations Development Programme Lake Tanganyika Biodiversity Project for
providing the bulk of the finances for this research.
Additional support for student involvement in the
project came from the US National Science Foundation (NSF Grant #s EAR 9510033 and ATM
9619458). We especially thank Drs. Andy Menz,
Graeme Patterson and Kelly West of the LTBP for
all of their gracious support at all stages of this
project, and the crew of the R/V Tanganyika
Explorer for their tireless efforts on our behalf
during the coring cruise. We gratefully acknowledge
the Tanzanian Council for Scientific Research
(COSTECH), the Tanzanian Fisheries Research
Institute (TAFIRI), and the University of Burundi,
for their support of this research program. We also
thank Dr. Patrick De Deckker and an unidentified
reviewer for their comments to improve this paper.
This is contribution #166 of the International
Decade of East African Lakes (IDEAL).
References
Alin S.R. and Cohen A.S. 2003. Lake level history of Lake
Tanganyika, East Africa, for the past 2500 years based on
ostracode-inferred water-depth reconstruction. Palaeogeogr.
Palaeoclimatol. Palaeoecol. 199: 31–49.
Alin S.R., O’Reilly C.M., Cohen A.S., Dettman D.L., PalaciosFest M.R. and McKee B.A. 2002. Effects of land-use change
on aquatic biodiversity: a view from the paleorecords at Lake
Tanganyika, East Africa. Geology 30: 1143–1146.
Bengtsson L. and Enell M. 1986. Chemical analysis, Chapter
21. In: Berglund B.E. (ed.), Handbook of Holocene Palaeoecology and Palaeohydrology. John Wiley and Sons,
Chichester, UK, pp. 423–451.
Bennett N.R. (ed.) 1970. Stanley’s Despatches to the New York
Herald, 1871–1872, 1874–1877. Boston University Press,
Boston, 508 pp.
Boyle J. 2004. A comparison of two methods for estimating the
organic matter content of sediments. J. Paleolimnol. 31: 125–
127.
Casanova J. and Hillaire-Marcel C. 1992. Late Holocene
hydrological history of Lake Tanganyika, East Africa, from
isotopic data on fossil stromatolites. Palaeogeogr. Palaeoclimatol. Palaeoecol. 91: 35–48.
Castañeda I., Erchak I. and Harper M. 1999. Total organic
matter, carbonate, and grain size determination in two eastwest coring transects, Luiche and Malagarasi River Deltas,
Lake Tanganyika, East African Rift Valley. In: Cohen A.
(ed.), The Nyanza Project 1999 Annual Report. Department
of Geosciences, University of Arizona, pp. 29–40.
Cohen A.S., Palacios-Fest M.R., McGill J., Swarzenski P.,
Verschuren D., Sinyinza R., Songori T., Kakagozo B.,
Syampila M., O’Reilly C.M. and Alin S.R. 2005a. Paleolimnological investigations of anthropogenic environmental
change in Lake Tanganyika: I. An introduction to the project. J. Paleolimnol. 34: 1–18.
Cohen A.S., Palacios-Fest M.R., Msaky E.S., Alin S.R.,
McKee B., O’Reilly C.M., Dettman D.L., Nkotagu H.H. and
Lezzar K.E. 2005b. Paleolimnological investigations of
anthropogenic environmental change in Lake Tanganyika:
IX. Summary of paleorecords of environmental change and
catchment deforestation at Lake Tanganyika and impacts on
the Lake Tanganyika ecosystem. J. Paleolimnol. 34: 125–145.
Cohen A.S., Talbot M.R., Awramik S.M., Dettman D.L. and
Abell P. 1997. Lake level and paleoenvironmental history of
Lake Tanganyika, Africa as inferred from late Holocene and
modern stromatolites. Geol. Soc. Amer. Bull. 109: 444–460.
Forester R.M. 1991. Ostracode assemblages from springs in the
western United States: implications for paleohydrology.
Mem. Entomol. Soc. Canada 155: 181–201.
49
Haberyan K.A. and Hecky R.E. 1987. The late Pleistocene and
Holocene stratigraphy and Paleolimnology of Lakes Kivu
and Tanganyika. Palaeogeogr. Palaeoclimatol. Palaeoecol.
61: 169–197.
Koponen J. 1988. People and Production in Late Precolonial
Tanzania, Vol. 2. Monogr. Finn. Soc. Development Studies,
434 pp.
McKee B., Cohen A.S., Dettman D.L., Palacios-Fest M.R.,
Alin S.R. and Ntungumburanye G. 2005. Paleolimnological investigations of anthropogenic change in Lake
Tanganyika: II. Geochronologies and mass sedimentation
rates based on 14C and 210Pb data. J. Paleolimnol. 34:
19–29.
Msaky E.S., Livingstone D. and Davis O.K. 2005.
Paleolimnological investigations of anthropogenic change
in Lake Tanganyika: V. Palynological evidence for
deforestation and increased erosion. J. Paleolimnol. 34:
73–83.
Nicholson S.E. 1999. Historical and modern fluctuations of
Lakes Tanganyika and Rukwa and their relationship to
rainfall variability. Clim. Change 41: 53–71.
Palacios-Fest M.R. 1994. Trace Element Shell Chemistry of
Continental Ostracodes and the Applicability of Experimentally-Derived Multiple Regression Models to Paleoenvironmental Reconstructions in Southwestern North America.
Ph.D. Dissertation, University of Arizona, 279 pp.
Stanley H.M. 1878. Through the Dark Continent v. 2. Harper
and Bros. Publishers, 566 pp.
Verschuren D. 2004. Decadal and century-scale climate variability in tropical Africa during the past 2000 years. In:
Battarbee R.W., Gasse F. and Stickley C. (eds), Past Climate
Variability through Europe and Africa. Developments in
Paleoenvironmental Research Vol. 6. Springer, Dordrecht.
Verschuren D., Laird K.R. and Cumming B.F. 2000. Rainfall
and drought in equatorial east Africa during the past
1,100 years. Nature 403: 41–414.